“Agents changed what one engineer can produce, and pretending otherwise is how we lose to the companies that adapt. The 212 PRs are real fixes customers wanted. If review is now the bottleneck, that is the next thing to fix, not a reason to produce less.”
212 PRs
A 10x engineer, and the nine people reviewing.
🧭 WHAT'S REALLY GOING ON
You've seen this when a leaderboard of merged PRs appears on a slide, and the people who reviewed them are not on it.
The real questionWhen one person's output goes up 10x, whose time paid for it?
⚖️ WHY BOTH ARE RIGHT
“Every PR costs a reviewer's attention, and someone has to understand the code we will maintain for ten years. My month went into reading 61 of Taylor's. The leaderboard counts where the work ended, not where it was done.”
🎯 SWEET SPOTS TO CONSIDER
Super Reasonable, the advisor who never takes a side
Measure team throughput
Celebrate what the team shipped, and show review time next to merge counts. If one person's output rises while everyone else's falls, the work moved; it did not multiply.
Make PRs cheap to review
Agent-made changes should arrive small, with the plan, the tests and a note on what to check. The author's job now includes making review fast.
Cap open PRs per author
A limit of five open PRs per person keeps output matched to the team's review capacity. A queue of 94 means slow down, not hire more reviewers.
Reward the multiplier
Put reviews, unblocked teammates and incidents prevented on the same slide as merges. In most teams, the 10x person is the one making nine others faster.
🚩 SIGNS YOU'VE GONE TOO FAR
- Taylor's side: you've overshot if agents open PRs faster than humans can read them, and approval has become a stamp to clear the queue.
- Ruth's side: you've overshot if every agent PR gets a line-by-line review, and the team's output is capped by its most careful reader.
🔬 IN THE FIELD GUIDE
Species observed in this story
CAST: WHO'S WHO
The team in this story
Same characters, same convictions. Learn their failure modes.
🤖 Storyboard for agentsLet’s make our agents LMFAO, or learn.
212 PRs
Premise: What does 10x output look like when review is shared?
- ALL-HANDS · TOP CONTRIBUTORS: Taylor · 212 PRs Everyone else: 2–14
- REVIEW LOAD: Ruth reviewed 61 of Taylor’s PRs Ruth’s own PRs: 2 · review queue: 94
Observed behavior: Some 10x output is somebody else’s month.
Cast: Maya Laurent · The Visionary · “Experiments beat roadmaps.”; Taylor Kim · The AI Native · “Give it to an agent.”; Ruth Lindqvist · The Maintainer · “We tried that in 2009.”
READ NEXT
